Re: tzcode or glib2 bug?

[christos%astron.com@localhost (Christos Zoulas)]

In article <20200210172958.azwyzrq3d773gm56@danbala>,
Thomas Klausner  <tk%giga.or.at@localhost> wrote:
>-=-=-=-=-=-
>
>Hi!
>
>Some weeks/months ago I noticed that some tools displayed the wrong
>time, one hour in the future (right now, on 9.99.46/amd64)
>
>I've tracked this down to those using glib2 and wrote a small test
>program.
>
># gcc test.c `pkg-config --cflags glib-2.0` `pkg-config --libs glib-2.0`
># ./a.out; date
>local time: 19:22:59
>UTC time: 17:22:59
>Mo. Feb. 10 18:22:59 CET 2020
>
>I have /etc/localtime symlinked:
>
>lrwxr-xr-x  1 root  wheel  33 May 31  2019 /etc/localtime ->
>/usr/share/zoneinfo/Europe/Vienna
>
>(and this is handled correctly by glib2) but it doesn't work with
>explicitly given TZ environment variable either.
>
># TZ=Europe/Vienna ./a.out; TZ=Europe/Vienna date
>local time: 19:24:45
>UTC time: 17:24:45
>Mo. Feb. 10 18:24:45 CET 2020
>
>Other timezones work:
>
># TZ=EST ./a.out; TZ=EST date
>local time: 12:24:35
>UTC time: 17:24:35
>Mo. Feb. 10 12:24:35 EST 2020
>
>Is there something weird in NetBSD's version of tzcode (I see we have
>quite some changes) and we notice because glib2 has its own parser, or
>is it a bug in glib2?
>
>(I debugged it into glib2 and found that it identifies the timezone
>correctly from the symlink and then calls init_zone_from_iana_info(),
>see
>https://github.com/GNOME/glib/blob/mainline/glib/gtimezone.c#L522.)

If you zic -b the zone does it work?

christos